German Court Decisions Scraper — Rechtsprechung im Internet (BGH, BVerfG, BVerwG, BFH, BAG, BSG, BPatG)

Get the full text of German federal court decisions as clean, structured JSON — straight from Rechtsprechung im Internet, the official corpus published by the German Federal Ministry of Justice and the Federal Office of Justice.

~84,000 decisions from seven federal courts, from January 2010 to last week.

Why this actor

• ✅ Official source, open data — the Bund publishes this corpus for free reuse. No protected sites, no fragile selectors, no legal gray zone.
• ✅ Real full text, not snippets — headnote (Leitsatz), tenor, facts (Tatbestand) and reasons (Gründe) as plain text, plus ECLI, cited norms and the previous instance.
• ✅ Full-text search over the whole corpus — the official site has no bulk search API; this actor gives you one.
• ✅ Monitor mode — schedule it and receive only decisions you have not seen yet. A daily case-law alert with zero noise.
• ✅ Built for RAG and AI agents — one flat JSON object per decision, ready to chunk and embed. Already anonymised at source: German court decisions are published pseudonymised.

Courts covered

Code	Court	Domain
BGH	Bundesgerichtshof	Civil & criminal — the largest share of the corpus
BVerfG	Bundesverfassungsgericht	Constitutional
BVerwG	Bundesverwaltungsgericht	Administrative
BFH	Bundesfinanzhof	Tax
BAG	Bundesarbeitsgericht	Labour
BSG	Bundessozialgericht	Social security
BPatG	Bundespatentgericht	Patents & trademarks

Typical use cases

Who	How
Legal-tech & LegalAI builders	Bulk-load a court's decisions into a vector store; ground your model in real German case law
Law firms	Daily monitor on BAG + Kündigung → new labour-law decisions in your inbox
Tax advisors	BFH decisions citing a specific norm (§ 15 UStG)
Patent attorneys	BPatG decisions on a trademark class or an opposition topic
Compliance & research	Track how a statute is applied over time — export to CSV/Excel and analyse

Input

Field	Example	Notes
courts	["BGH", "BAG"]	Empty = all seven federal courts.
dateFrom / dateTo	"2026-01-01"	Decision date, not publication date. Corpus starts 2010-01-04.
keywords	Kündigung Betriebsrat OR "fristlose Kündigung"	Full-text. AND between terms, OR between groups, "quotes" for phrases. Umlaut-insensitive (Kundigung finds Kündigung).
caseNumber	"IX ZB 72/08"	Substring match on the Aktenzeichen.
docTypes	["Urteil"]	Urteil, Beschluss, Gerichtsbescheid…
includeFullText	true	Off = lightweight citation index (fast, cheap).
monitorMode	true	Only decisions not returned by a previous run with the same filters.
maxResults	50	Safety cap.
maxScanned	600	How deep a keyword search digs before stopping.

Output example

{
    "docId": "JURE100054597",
    "ecli": "ECLI:DE:BVERWG:2010:080110B9B3.09.0",
    "court": "BVerwG",
    "courtBody": "9. Senat",
    "decisionDate": "2010-01-08",
    "caseNumber": "9 B 3/09",
    "docType": "Beschluss",
    "norms": ["§ 133 Abs 3 S 1 VwGO"],
    "title": "Verwerfung der Nichtzulassungsbeschwerde",
    "guidingPrinciple": null,
    "previousInstance": "vorgehend Oberverwaltungsgericht des Landes Sachsen-Anhalt, 2. September 2008, Az: 4 L 572/04, Urteil",
    "tenor": "Die Beschwerde des Klägers gegen die Nichtzulassung der Revision … wird verworfen.",
    "reasons": "Die Beschwerde ist als unzulässig zu verwerfen. Sie ist nicht innerhalb von zwei Monaten …",
    "sourceUrl": "https://www.rechtsprechung-im-internet.de/jportal/?quelle=jlink&docid=JURE100054597&psml=bsjrsprod.psml&max=true",
    "source": "rechtsprechung-im-internet.de"
}

Every run also writes a RUN_SUMMARY record to the key-value store — result counts, how many documents were downloaded, and the resolved filters. Handy for integrations, debugging and AI agents.

Scheduling a case-law alert

1. Set courts + keywords, enable monitorMode.
2. Create a Schedule in Apify Console (e.g. every weekday at 7:00).
3. Add an integration (email, Slack, webhook, Make, Zapier) — you are only notified when something new lands.

Set minExpectedResults on a broad scheduled run to be alerted if the upstream corpus ever stops updating.

Pricing

Pay per event: a small fee per run start plus a fee per decision returned. A monitor run that finds nothing new costs almost nothing.

A note on the connection

The corpus is served from a German edge that refuses connections from most cloud datacenters, so the actor routes its requests through Apify Proxy. The default (Apify Proxy, automatic) works out of the box — you do not need to configure anything, and you do not need a specific proxy country.

Data source & fair use

All data comes from rechtsprechung-im-internet.de, operated by the Bundesministerium der Justiz and the Bundesamt für Justiz, and offered for free reuse. Decisions are published in pseudonymised form by the courts themselves — this actor adds no personal data and performs no re-identification. Downloads are rate-limited and run at a low, polite concurrency.

Roadmap

• Austrian (RIS) and Czech (NSS / ÚS) case law through the same output schema
• Norm-level filtering (§ 15 UStG) as a first-class input
• Landesgerichte via openJur